Defocused Blur Image Restoration Using Cepstral Correlation and Wiener Filtering in MATLAB

For defocused blur images, we have implemented a MATLAB algorithm that initially estimates Point Spread Function (PSF) parameters based on cepstral correlation analysis, then performs image restoration using Wiener filtering. This approach significantly improves the quality of defocused blur images, making them sharper and more recognizable. The implementation involves two key computational stages: First, the cepstral correlation method analyzes the logarithmic power spectrum to identify PSF characteristics, where MATLAB's frequency domain functions (fft2, ifft2) are utilized to compute the cepstrum and detect blur parameters. Second, the Wiener filter applies frequency-domain restoration using estimated PSF parameters, incorporating noise-to-signal ratio considerations for optimal deblurring. The algorithm effectively handles typical defocus blur patterns and provides substantial improvement in image clarity and detail recognition. Key MATLAB functions employed include fft2/ifft2 for Fourier transformations, deconvwnr for Wiener filtering implementation, and custom correlation analysis functions for PSF parameter extraction. The method demonstrates practical effectiveness for enhancing defocused images in various computer vision and image processing applications.